a general mathematical model for analyzing the performance
TRANSCRIPT
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
1/14
A general mathematical model for analyzing the performanceof fuel-cell membrane-electrode assemblies
Huayang Zhu*, Robert J. KeeEngineering Division, Colorado School of Mines, Golden, CO 80401, USA
Received 9 January 2003; accepted 27 January 2003
Abstract
We have developed a general mathematical model to represent the membrane-electrode assembly (MEA) of fuel-cell systems. The model is
used to analyze the effects of various polarization resistances on cell performance. The model accommodates arbitrary gas mixtures on theanode and cathode sides of the MEA. Moreover, it accommodates a variety of porous electrode and electrolyte structures. Concentration
overpotentials are based on a dusty-gas representation of transport through porous electrodes. The activation overpotentials are represented
using the ButlerVolmer equation. Although the model is general, the emphasis in thispaper is on solid-oxide fuel-cell (SOFC) systems for the
direct electrochemical oxidation (DECO) of hydrocarbons.
# 2003 Elsevier Science B.V. All rights reserved.
Keywords: Model; MEA; Fuel cell; SOFC
1. Introduction
Quantitative models can be valuable in the interpretation
of experimental observations and in the development andoptimization of systems. The objective of the membrane-
electrode assembly (MEA) model developed herein is to
provide a physically based model that can be used to
evaluate the effects of design and operating alternatives.
While the model can stand alone, the software is written in a
way that enables incorporation into larger system-level
models that also consider fluid flow and thermal manage-
ment throughout a fuel-cell system. The model is general in
the sense that it can accommodate a variety of geometrical
configurations. It can also represent proton-conducting or
oxygen-ion-conducting electrolytes. The discussion and
examples in this paper, however, consider solid-oxide
fuel-cell (SOFC) system.
The great advantage of SOFC systems for highly efficient
electric power generation lies in its potential for direct use of
hydrocarbon fuels, without the requirement for upstream
fuel preparation, such as reforming [14]. A typical layout
for a planar design of the SOFC system is illustrated in Fig. 1.
The membrane-electrode assembly, which is composed of
an electrolyte sandwiched between the anode and the cath-
ode, is itself sandwiched between metal interconnect struc-
tures. The fuel and air channels are formed in the
interconnect structure. The electrolyte is a dense ceramic
(e.g. yttria-stabilized zirconia (YSZ) or gadolinia-doped
ceria (GDC)), which is impermeable to gas flow but is anoxygen-ion (O2) conductor. The composite electrodes are
porous metal-loaded ceramics (cermets) (e.g. NiYSZ for
the anode and Sr-doped LaMnO3 (LSM) for the cathode),
which may be mixed electronic and ionic conductors that
can extend the three-phase boundaries (TPB) and promote
the electrocatalytic reactions.
The SOFC system involves complex transport, chemical,
and electrochemical processes, with its operating perfor-
mance strongly affected by the corresponding transport
resistances and the activation barriers. Fig. 2 illustrates some
of these processes at the MEA level. During operation,
oxygen from the air channel transports through the porous
cathode, whereupon it is reduced to oxygen ion (O2) at
gascathodeelectrolyte three-phase boundaries. The
formed oxygen ion at the cathodeelectrolyte interface is
then transported to the anodeelectrolyte interface through
the ion-conducting electrolyte. At the same time, fuel from
the fuel channel transports through the porous anode to the
anodeelectrolyte interface, where it is oxidized electroche-
mically at the gasanodeelectrolyte three-phase boundaries
to products. The products (e.g. H2O and CO2) are then
transported back to the fuel channel through the porous
anode structure. Transport resistances of the gas-phase
species in the porous electrodes and O2 in the electrolyte,
Journal of Power Sources 117 (2003) 6174
* Corresponding author. Tel.: 1-303-273-3890; fax: 1-303-273-3602.E-mail address: [email protected] (H. Zhu).
0378-7753/03/$ see front matter # 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0378-7753(03)00358-6
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
2/14
as well as the activation energy barriers for electrochemical
reactions can result in various polarizations. They are repre-
sented as concentration overpotentials, activation overpo-
tentials, and ohmic overpotentials. The main contribution to
the ohmic polarization is from the transport resistance of
O2 in the electrolyte. The concentration polarization is due
to the transport resistance of the gaseous species through
porous composite electrodes, and the activation polarization
is related to the charge-transfer processes at the electrode
electrolyte interfaces. These various polarizations are func-
tions of both operating conditions and the physical proper-
ties of the cell. The operating conditions include
temperature, pressure, and fuel and oxidizer concentrations.
The cell properties involve materials, macro- and micro-structures of the electrolyte and the composite electrodes,
which include porosity, tortuosity, permeability, and thick-
ness of the anode and the cathode, the ionic conductivity and
thickness of the electrolyte, the active area and activity of the
electrodeelectrolyte interface.
To understand these various resistances and enhance the
cell performance, parametric studies for these various polar-
izations in the electrolyte and composite electrodes have
been widely made through experiments [5,6], theoretical
analysis [7], and numerical models [810]. The theoretical
analyses and numerical models for the activation polariza-
tions require an understanding the elementary thermal
electrochemical reaction mechanisms and the microstruc-
ture of the electrode. Due to the complex and uncertain
details of the reaction processes (particularly in case of
hydrocarbons), most modeling in the literature has focused
on using H2 as the fuel. Chan et al. [8] used the intrinsic
charge-transfer resistance to represent the activation polar-
ization. Tanner et al. [11] have derived an effective charge-
transfer resistance for the composite electrode, which is a
function of the microstructural parameters of the electrode,
intrinsic charge-transfer resistance, ionic conductivity of theelectrolyte, and the electrode thickness. This effective
charge-transfer resistance model was later used to analyze
the activation overpotentials [9,10]. Adler et al. [12] derived
a charge-transfer resistance model for the composite cath-
ode, which is a function of the tortuosity, porosity, interface
area, oxygen self-diffusion coefficient and oxygen exchange
coefficient. The activation polarizations for a particular
electrode may also be measured by the electrochemical
impedance spectroscopy (EIS) [5,6,13]. Previous porous-
media gas-phase transport models have been developed to
predict concentration overpotentials [810]. These include
molecular diffusion, Knudsen diffusion, and pressure-driven
Darcy flow. However, these have only been applied in fuel-
cell systems with binary gas-phase mixtures.
The model developed in this paper represents a unit-cell
structure (Fig. 2), considering the effects of various over-
potentials on the cell performance. As illustrated in Fig. 2
the cell is anode-supported, with a very thin electrolyte and
cathode, and both electrodes are porous ceramicmetal
composite structures. Channels formed in an interconnect
structure carry the flow of fuel and air. Depending on the
position of the unit cell within a fuel-cell system, the
composition in both the fuel and air channels varies depend-
ing on initial composition and depletion.
Fig. 1. Typical layout of a planar SOFC. The left-hand panel illustrates the system at the scale of flow channels (in mm), while the right-hand panel illustrates
the small-scale structure (in mm) of the porous cermet electrode.
Fig. 2. Illustration of the components of the MEA unit cell.
62 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
3/14
2. Model formulation
The overall performance of SOFC systems is greatly
affected by ohmic, concentration, and activation polariza-
tions. Thus, the operating cell potential (Ecell) is lower than
the Nernst potential (E). The operating cell potential (Ecell)
can be formally expressed as:
Ecell E Zconc;a Zact;a Zohm Zconc;c Zact;c
Zinterface Zleakage; (1)
where Zconc;a and Zconc;c are the concentration overpotentials
at the anode and the cathode due to the gas-phase species
diffusion resistance, Zact;a and Zact;c the corresponding acti-
vation overpotentials, Zohm the ohmic overpotential in the
electrolyte, Zinterface the interface overpotential due to the
contact resistance at material boundaries, andZleakage the cell
potential loss due to the electric current through the electro-
lyte. Subsequent sections in the paper develop models for
each of these overpotentials, all of which are functions of thecurrent density ie.
2.1. Global electrochemical reaction
A global electrochemical reaction can be written as:
nf0XKk1
nf;kwk
! no
0XKk1
no;kwk
!?
XKk1
n00f;kwk XKk1
n00o;kwk;
(2)
where wk is the chemical symbol for the kth species (which
may participate as a fuel, an oxidizer, or both),nf0 and no
0 the
stoichiometric coefficients for the fuel and oxidizer mix-tures, and n00f;k and n
00o;k the stoichiometric coefficients of
the kth product species in the fuel and air channels, res-
pectively. Each species wk has a primary identity as a fuel,
an oxidizer, a product, or an inert, and this identity plays
an essential role in balancing the reaction to determine
stoichiometric coefficients and in assigning charge trans-
fer. The global reaction can be expressed in compact form
as:
XK
k1
nk0wk?
XK
k1
n00kwk; (3)
where nk0 nf
0nf;k no0no;k and n
00k n
00f;k n
00o;k.
Fuel and oxidizer mixtures are separated by the anode
electrolytecathode MEA structure. Thus, in writing a glo-
bal electrochemical reaction and calculating the Nernst
potential, it is convenient to represent the fuel mixture
and the oxidizer mixture separately as reactants. This
separation is accomplished by writing the LHS of the
reaction as two summations. The fuel channel mixture
composition is represented in terms of the species mole
numbers as nf;k. Similarly, no;k is used to represent the
mixture of the composition of the oxidizer mixture in the
air channel (e.g. for air, no;O2 0:21 and no;N2 0:79).
The RHS of the global reaction is also separated into two
terms. Depending on whether the cell is an oxygen-ion
conductor or a proton conductor, the product species appear
on the anode or cathode side of the MEA. Thus, following
reaction balancing, the location of the product species
affects the product stoichiometric coefficients n00f;k and
n00o;k. There is more discussion on this point following the
discussion on reaction balancing.
Since the electrochemical reactions involve charge trans-
fer it is necessary to specify the charge transfer associated
with each reactant species. For this purpose, we introduce
two sets of parameters (zf;k and zo;k), which specify the
charge transferred per mole of the kth fuel and oxidizer
species. For a species that is inert relative to the electro-
chemical charge-transfer process, z 0. In our convention,the product species (e.g. H2O and CO2) also are assigned
z 0. Consider two examples for assigning z. Since O2 inthe air channel is reduced to O2 by obtaining four electrons
(i.e. O2 4e ! 2O2), the parameter zo;O2 4. Because
the direct electrochemical reaction of CH4 with O2 at the
anode to produce H2O and CO2 (i.e. CH4 4O2 !
CO2 2H2O 8e) releases eight electrons, zf;CH4 8.
The total number of electrons transferred by the global
electrochemical reaction is represented as:
ne XKk1
nf0nf;kzf;k
XKk1
no0no;kzo;k: (4)
There may be reasons to incorporate oxidizer species into
the fuel channel or fuel species into the oxidizer channel. For
example, oxygen in the fuel channel may promote beneficial
partial oxidation of the fuel or it may work to inhibit depositformation. Such species, while influencing thermal chem-
istry in the gas or on catalytic surfaces, do not participate
directly in the electrochemical reaction. Thus, they are
considered inert in the global electrochemical reaction.
For example, if O2 (i.e. an oxidizer species) appears as part
of the fuel composition (i.e. in the first term of the global
reaction (2)), then zf;O2 0. The added oxygen in the fuelchannel cannot work directly to produce electric energy
only fuel oxidization with oxidizer from the cathode channel
can produce power. Thus, any oxidizer in the fuel channel
must reduce the cell potential because of its diluting effect.
Before proceeding to evaluate the cell potential, the global
reaction must be balanced to determine the stoichiometric
coefficients. Although balancing reactions is a well known
process, a few words of explanation are warranted. A
balance equation is written for each element (e.g. H, C,
O, N), requiring that the reaction conserves the element
balance. The process leads to a system of linear equations for
the stoichiometric coefficients (nf0, no
0, and n00k). Because of
an essential linear dependence stemming from the fact that
species are composed of elements in a specified way, one can
arbitrarily set nf0 1. Then depending on the number of
product species, only some of the n00k are non-trivial. Spe-
cifically, the number of product species (and hence the
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 63
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
4/14
number n00k coefficients) is one fewer than the number of
elements. Once the overall product stoichiometric coeffi-
cients n00k are known, then either:
n00f;k n00k or n
00o;k n
00k; (5)
depending on the type of cell (i.e. oxygen ion or proton
conducting) and hence the location of the product species.Inert species (i.e. reactantspecies for which zf;k 0 and
zo;k 0) are initially excluded from the balancing proce-dure. Once the balance is completed, the inert species are
reintroduced with the same stoichiometric coefficients on
the reactant and product sides of the global reaction. The
entire balancing process is completely automated in the
software.
2.2. Nernst potential and the Nernst equation
Based on the chemical potential balance at open circuit,
the Nernst equation provides that:
E E RT
neF
XKk1
nf0nf;k no
0no;k n00k ln
pk
p0
; (6)
where E is the Nernst electric potential, E the ideal Nernst
potential at standard conditions (p0 1 atm), pk the partialpressure of the kth species, T the temperature, R 8:314 J/(mol K) is the universal gas constant, and F 96485:309 C/mol is Faradays constant. The ideal Nernst potential at the
standard conditions is given as:
E DG
neF; (7)
where DG is the change in standard-state Gibbs free energy
between products and reactants of the global reaction Eq. (3).
Specifically:
DG XKk1
nf0nf;k no
0no;k n00km
k; (8)
where mk is the standard-state chemical potential of the kth
species. The standard-state thermodynamic properties of
ideal gases depend on temperature. The needed thermody-
namic properties are readily available in databases such as in
CHEMKIN [14]. As an alternative to using the species partial
pressures, the Nernst equation can also be represented interms of the species molar concentrations Xk as:
E E RT
neF
XKk1
nf0nf;k no
0no;k n00k lnXk
RT
neFln
RT
p0
XKk1
nf0nf;k no
0no;k n00k: (9)
Fig. 3 illustrates the Nernst potentials for three fuels as a
function of fuel utilization (at T 750 8C and p 1 atm).Consider the situation for butane where the global reaction is:
C4H10 132
O2 ! 5H2O 4CO2: (10)
If the fuel is 50% utilized then the fuel mixture is composed
of 1/2 moles of C4H10, 5/2 moles of H2O, and 4/2 moles of
CO2. The Nernst potentials are highest for the hydrocarbons,
assuming direct electrochemical oxidation. The figure also
shows that the Nernst potential for hydrogen is reduced most
as a function of fuel depletion. The potentials illustrated in
Fig. 3 are calculated presuming that the oxidizer is pure air
(i.e. no depletion of the oxygen in the air).
2.3. Concentration polarization
For open-circuit conditions (i.e. zero current flow), thespecies concentrations at the electrolyte interface (three-
phase boundary) are the same as those in the bulk channel
flow. Denoting the molar species concentrations in the
channels as Xk
, the Nernst potential is written as:
E E RT
neF
XKk1
nf0nf;k no
0no;k n00k lnXk
RT
neFln
RT
p0
XKk1
nf0nf;k no
0no;k n00k: (11)
However, when the current is flowing there must be con-
centration gradients across the electrode structures. This isbecause the diffusion processes that supply the species flux
to the electrochemical reactions are driven by concentration
gradients. Consequently, species concentrations at the three-
phase boundaries Xks
are different from the bulk concen-
trations. Evaluated at the three-phase boundaries, the corre-
sponding Nernst equation is written as:
Es E RT
neF
XKk1
nf0nf;k no
0no;k n00k lnXk
s
RT
neFln
RT
P0
X
K
k1
nf0nf;k no
0no;k n00k: (12)
Fig. 3. Nernst potential for three fuels-air systems as a function of
percentage of the fuel utilization. As the fuel is utilized it is converted to
stoichiometric products that dilute the fuel on the anode side. The air is not
depleted in this system.
64 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
5/14
The potential difference associated with the concentration
variation is written in terms of a concentration overpotential
as follows:
Zconc E Es
RT
neFX
K
k1
nf0nf;k no
0no;k n00
f;k n00
o;k ln
Xk
Xks :
(13)
The equation can easily be rewritten to separate the anode
and cathode contributions as:
Zconc RT
neF
XKk1
nf0nf;k n
00f;k ln
Xk
Xks
RT
neF
XKk1
no0no;k n
00o;k ln
Xk
Xks
: (14)
Furthermore, the concentration overpotentials at the anode
and the cathode can be defined separately as:
Zconc;a RT
neF
XKk1
nf0nf;k n
00f;k ln
Xk
Xks
; (15)
Zconc;c RT
neF
XKk1
no0no;k n
00o;k ln
Xk
Xks
: (16)
There are a number of physical processes that may con-
tribute to the concentration polarization. These include gas
species molecular transport in the electrode pores, solution
of reactants into the electrolyte and dissolution of products
out of the electrolyte, and the diffusion of the reactants/
products through the electrolyte to/from the electrochemicalreaction sites. Typically, however, the diffusive transport of
the reactants and products between the flow channels and the
electrochemical reaction sites across the electrode is the
major contributor to the concentration polarization. For an
anode-supported MEA structure, the concentration polari-
zation at the cathode is usually very small due to its being
very thin (e.g. 50 mm). However, the concentration polar-
ization at the anode can be very significant, particularly at
the high current density and high fuel utilization rate.
2.3.1. Dusty-gas model
The concentration polarization is caused primarily by the
transport of gaseous species through porous electrodes. It is
affected by the microstructure of the electrodes, particularly,
the porosity, permeability, pore size, and tortuosity factor.
There are four major physical processes that affect transport
in the porous media. They are molecular diffusion, Knudsen
diffusion, surface diffusion, and Darcys viscous flow. The
relative importance of bulk diffusion (e.g. Fickian) and
Knudsen diffusion depends on the relative frequency of
molecular-molecular collisions and molecular-wall colli-
sions. These processes can be characterized by the Knudsen
number Kn l=d, where l is the mean free path length inthe gas and d is a characteristic pore diameter. When
Kn ( 1 bulk diffusion is dominant and when Kn ) 1Knudsen diffusion is dominant. When Kn % 1, which isthe typical situation in SOFC electrodes, both bulk diffusion
and Knudsen diffusion are comparable and must be con-
sidered together. The dusty-gas model (DGM) [15] was
developed to represent multicomponent transport in this
situation.The dusty-gas model can be used to describe the relation-
ship between the gas species concentrations in the flow
channel and concentrations at the electrochemical reaction
sites (at the electrodeelectrolyte interface). The DGM was
developed initially for astrophysical applications, assuming
that the pore wall consists of giant, motionless, and uni-
formly distributed particles (i.e. dust) in space. Three gas-
phase transport mechanisms are considered: molecular dif-
fusion, Knudsen diffusion, and viscous flow. Bulk diffusion
and Knudsen diffusion are combined in series to form the
total diffusive flux. The viscous porous-media flow (Darcy
flow) acts in parallel with the diffusive flux. There are three
porous-media parameters that appear in the DGM. They are
the Knudsen coefficient (Kg), the permeability constant (Bg),
and the porosity-to-tortuosity ratio (fg=tg), all which can bedetermined experimentally. The DGM can be written as an
implicit relationship among the molar concentrations, molar
fluxes, concentrations gradients, and the pressure gradient
[16,17]:
X6k
XNg
k XkNg
XTDek
Ng
k
Dek;Kn
rXk Xk
De
k;Kn
Bg
mrp: (17)
In this relationship Ng
k is the molar flux of species k, Xk themolar concentrations, and XT p=RT is the total molarconcentration. The mixture viscosity is given as m (which
depends on the mixture composition and temperature) and
Dek and Dek;Kn are the effective molecular binary diffusion
coefficients and Knudsen diffusion coefficients.
The effective molecular binary diffusion coefficients in
the porous media Dek are related to the ordinary binary
diffusion coefficients Dk in the bulk phase [18] as:
Dek fg
tg
Dk: (18)
Binary diffusion coefficients, which are determined from
kinetic theory, may be evaluated with software such as
CHEMKIN [18]. Knudsen diffusion, which occurs due to
gaswall collisions, becomes dominant when the mean free
path of the molecular species is much larger than the pore
diameter. The effective Knudsen diffusion coefficient can be
expressed as:
Dek;Kn 4
3Kg
ffiffiffiffiffiffiffiffiffi8RT
pWk
r; (19)
where the Knudsen permeability coefficient Kg rpfg=tg
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 65
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
6/14
(fg is the porosity and tg the tortuosity factor), Wk are the
molecular weights, and rp the average pore radius.
Assuming that the porous electrode is formed by closely
packed spherical particles with diameter dp (an idealization),
the permeability can be expressed by the KozenyCarman
relationship as [19]:
Bg f3gd
2p
72tg1 fg2: (20)
Other porous-media situations have different permeabilities.
An iterative procedure is needed to determine the pressure
and concentration gradients. Assuming that the species
molar fluxes Ngk through the porous electrode are known
based on the current density and the global electrochemical
reaction at the electrodeelectrolyte interface, the pressure
gradient can be determined as:
rp PkN
g
k=Dek;Kn
1=RT Bg=mP
kXk=De
k;Kn
: (21)
This expression for the pressure gradient comes from a
summation of Eq. (17) over all species k, recognizing that
summation of the first term (representing ordinary diffu-
sion) vanishes exactly. The concentrations Xk in thedenominator are evaluated either in the channel or at the
electrolyte interface, depending on the direction of the net
molar fluxP
k Ng
k. If the net flux is toward the electrolyte (as
it normally is in the cathode), then the Xk are taken as thechannel concentrations. However, when the flux is toward
the channel (as it usually is in the anode), the concentrations
are evaluated at the electrolyte interface. The iteration is
needed because the electrolyte interface concentrations arenot yet known. Assuming an initial guess for the concen-
trations Xk (usually the channel concentrations), the pres-sure gradient is evaluated. Then, using this pressure
gradient, the concentration gradients can be determined
from Eq. (17) as:
rXk X6k
XNg
k XkNg
XTDek
Ng
k
Dek;Kn
Xk
Dek;Kn
Bg
mrp: (22)
The concentrations in the summation on the RHS are
evaluated as discussed above depending on the direction
of the net flux. This analysis is simplified greatly by assum-
ing linear profiles of pressure and concentrations across the
electrode. We have justified this assumption by solving the
two-dimensional dusty-gas model over a large range of
electrode structures and fuel-cell operating conditions. Once
the concentration gradients are evaluated, a new estimate for
the concentrations at the TPB can be found (assuming linear
concentration profiles and a known electrode thickness).
With the new estimated concentrations a new pressure
gradient can be found. This iteration converges very
rapidlytypically needing only two or three iterations.
Assuming a current density ie and a global electroche-
mical reaction, the molar flux of the gas species through the
anode and cathode can be evaluated as:
Ng
k;a nf0nf;k n
00f;k
ie
neF; (23)
Ngk;c no 0no;k n00o;k ie
neF: (24)
Note that the molar flux must be taken as zero for any inert
species that does not participate in the electrochemical
reactions (e.g. nitrogen in the cathode). Appendix A pro-
vides examples.
2.4. Activation polarization and the ButlerVolmer
equation
In the chemical reaction process, the reactants must usually
overcome an energy barrier, that is, the activation energy. In
charge-transfer electrochemical reactions, the reactants mustovercomenot only a thermal energybarrier but also an electric
potential. Thus, a portion of the potential cell voltage is lost
in driving the electrochemical reactions that transfer the
electrons to or from the electrodes. This activation polariza-
tion is a very important loss when the electrochemical reac-
tions are controlled by the slow electrode kinetics.
Assuming a single rate-controlling reaction, the Butler
Volmer equation [20] provides a relationship between the
current density and the charge-transfer overpotential as:
ie i0 exp aanBVe FZact
RT exp acnBVe FZact
RT : (25)This equation represents the net anodic and cathodic current
due to an electrochemical reaction. The activation over-
potential Zact is potential difference above the equilibrium
electric potential between the electrode and the electrolyte.
The factor nBVe is the number of electrons transferred in the
single elementary rate-limiting step that the ButlerVolmer
equation represents. It is quite common to assume nBVe 1.The anodic and cathodic asymmetric factors (also called
charge-transfer coefficients) aa and ac determine the relative
variations of the anodic and cathodic branches of the total
current with an applied overpotential Zact. The anodic and
cathodic charge-transfer coefficients aa
and ac
depend on the
electrocatalytic reaction mechanism and typically take
values between zero and one. Further, they are constrained
by aa ac 1. The exchange current density i0 is thecurrent density of the charge-transfer reaction at the
dynamic equilibrium electric potential difference Eeq. In
other words, at the equilibrium potential, the forward and
reverse current densities are equal at i0. A high exchange
current density implies the reaction proceeds rapidly upon
varying the potential from its equilibrium value. There is an
activation overpotential for the anode and the cathode.
Consequently, there are charge-transfer coefficients aa and
ac for both the anode and the cathode.
66 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
7/14
There are two potentially interesting limits of the Butler
Volmer equation. At very high surface activation overpo-
tential one of the exponential terms in ButlerVolmer equa-
tion becomes negligible. The resulting Tafel equation is a
straight line on a semi-logarithmic plot, lnie $ Zact. Alter-natively stated:
Zact RT
aanBVe F lnie lni0 for
aanBVe FZactRT
) 1;
(26)
or
Zact RT
acnBVe F lnie lni0
for acn
BVe FZactRT
) 1: (27)
In an SOFC, the Tafel equation can often be used to calculate
the activation overpotential at the cathode. The second
limiting case is realized at very lower activation polariza-tion. In this case, the following linear current overpotential
relationship applies:
Zact RT
aa acneF
ie
i0
: (28)
This equation can often be applied to calculate the activation
potential at the anode of the SOFC system. In our model
neither of these limiting cases is actually used to determine
the activation overpotentials. Rather, the ButlerVolmer
equation is solved iteratively in its native form.
The exchange current density i0 is a measure of the
electrocatalytic activity of the electrodeelectrolyte inter-face or TPB for a given electrochemical reaction. In fact,
the exchange current density is a crucially important factor
for determining the activation overpotential Zact. However,
i0 is not a simple constant parameter, but its value may
depend on the operating conditions and material properties,
including concentrations of reactants and products adjacent
to the electrode, temperature, pressure, the microstructure
and electrocatalytic activity of the electrode, and even
the conductivity of the electrolyte [21]. Particularly for
composite SOFC electrodes with the mixed ionic and
electronic conduction, such as a NiYSZ for the anode
and a LSMYSZ for the cathode, the electrochemical
reaction zone can be extended from the electrolyte-elec-
trode interface into the electrode. Thus, the charge-transfer
resistance can be reduced and the exchange current density
can be enhanced. Furthermore, the exchange current den-
sity for the composite electrode becomes a function of the
electrochemical properties of the electrolyte and electrode
and the effective TPB length. Thus, microstructural proper-
ties of the composite electrode, such as the grain size and
volume fraction of the electrolyte, will affect the exchange
current density. A steady-state currentvoltage experiment
can measure i0 at a particular operating condition. How-
ever, it is very difficult to obtain a general analytical
expression for i0, which requires detailed understanding
of the elementary thermal and electrochemical reaction
mechanisms occurring at the electrode and the microstruc-
ture of the electrode.
In our model we provide flexibility for the exchange
current density to depend on the reactant concentrations
as:
i0 i00
YKk1
Xgkk ; (29)
where i00 is a constant and gk and Xk the reaction order and
mole fraction (at the electrodeelectrolyte interface) for the
kth species, respectively. Although this formulation provides
flexibility in the model, it is still far from a general descrip-
tion of the elementary charge-transfer reactions steps.
Furthermore, it is difficult to determine the reaction orders
for any particular system.
At the cathode of an SOFC system, the gas-phase oxygen
reduction and ion incorporation into the electrolyte is a
complex process that involves a series of elementary reac-
tions steps, including the oxygen adsorption and dissociation
on the surface, diffusion of the intermediate species on
the surface, charge transfer, and oxygen incorporation into
the lattice of the electrolyte. Any of the reaction steps may
be the rate-limiting reaction for the overall process, depend-
ing on the operating conditions (temperature, oxygen partial
pressure), and material properties, and microstructure of
the electrode (catalytic activity, pore size, porosity, tortuos-
ity, and permeability). For the LSMYSZ composite cath-
ode, Kim et al. [22] have indicated that: (1) i0 $ p3=8O2
if
the rate-determining step is the charging step of adsorbedoxygen; (2) i0 $ p
1=4O2
if the rate-determining step is diffusion
of charged surface oxygen to the TPB; and (3) i0 does not
depend on pO2 if the rate-determining step is incorporation
of oxygen ions into the electrolyte. For the PtYSZ cathode,
the exchange current density i0 was shown to be proportional
to p1=4O2
at low pO2 and p1=4O2
at high pO2 [23]. Uchida et al.
[21] indicated that i0 may be proportional to the ionic
conductivity of the electrolyte for both the LSMYSZ
and PtYSZ cathodes depending on the operating tempera-
tures.
The situation is more complex at the anode of an SOFC
system. Reaction of a hydrocarbon fuel may involve direct
electrochemical oxidization [24], steam and/or dry reform-
ing, pyrolysis, and deposit formation. The general thermal
and electrochemical reaction steps may include dissocia-
tive adsorption on surfaces, surface and/or charge-transfer
reactions among the adsorbed intermediate species, surface
and/or bulk diffusion of the adsorbed species, and deso-
rption of products. Even for hydrogen, the simplest and
most widely studied fuel, there is still considerable debate
about the underlying electrochemical processes [5,7,25
27]. The rate-determining reaction step of the hydrogen
oxidation at the anode may be the dissociative adsorption
of hydrogen, the formation of hydroxyl, a charge-transfer
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 67
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
8/14
reaction, or the desorption of water, all of which can
depend on the operating conditions (such as the tempera-
ture, pressure, and species concentrations), macro- and
microstructures, and material properties of the electrode.
Additionally thermal and electrochemical reactions may
occur on the surface and in the bulk of the metal and/or
ceramic of the composite cermet electrode. Thus, there isnot a simple means to represent the relationship between
the exchange current density and the species concentra-
tions at the anode.
There are ways to measure the exchange current density
(or infer its value from related measurements). Based on the
ButlerVolmer equation, a charge-transfer resistance Rct can
be expressed as:
R1ct Atpb@ie@Z
Xk;T
i0AtpbnBVe F
RT
aa exp aanBVe FZ
RT
ac exp acnBVe FZ
RT
: (30)
At the zero activation overpotential:
Rct RT
i0Atpbaa acnBVe F; (31)
which presents a relationship between the exchange current
density i0 and the charge-transfer resistance Rct. The charge-
transfer resistance can be determined from EIS measure-
ments [6,13].
2.5. Ohmic polarization
Sources of ohmic losses in a fuel cell are the resistance to
the ion flow in the electrolyte and the resistance to the
electronic flow in the electrode. Due to typically high metal
loading resistive electric conductivity in the electrodes is
high, which limits ohmic losses in the electrodes. Thus,
ohmic polarization in an SOFC system is typically domi-
nated by ion resistance through the electrolyte. Such losses
can be reduced by decreasing the electrolyte thickness and
enhancing its ionic conductivity.
Since electric current flow in the electrolyte and electro-
des obey Ohms law, the ohmic losses can be expressed
simply as:
Zohm ieRtot; (32)
where Rtot is the total area-specific cell resistance, including
the area-specific resistances in the electrodeRed and the solid
electrolyte Rel. The ionic conductivity of the electrolyte can
usually expressed as:
sel s0T1 exp
Eel
RT
; (33)
where Eel is the activation energy for the ionic transport.
Thus, the specific ohmic resistance of the electrolyte with
thickness of Lel can be obtained as:
Rel Lel
sel: (34)
There are several opportunities for interface overpoten-
tials with in the MEA structure. For example, anywhere
two materials are physically bonded together there canbe a resistance the causes an interface overpotential,
Zinterface.
If the electrolyte has electrical conductivity, there is a
possibility for a leakage current that may be represented as
a leakage overpotential Zleakage. In our model:
Zleakage Z0leakage 1
i
imax
; (35)
where Z0leakage is the leakage overpotential at open circuit and
imax the maximum current density for the cell. Once the
Nernst potential is known, Z0leakage can be observed from
measurement of the open-circuit potential. The value of imaxcan be determined from the model by calculating current
density at which the cell power density vanishes.
3. Solution algorithm
The model is solved by determining the cell voltage for
a given current density, or vice versa. In general, the cell
potential Ecell is expressed as the difference between the
Nernst cell potential E and the sum of all the relevant
overpotentials, which depend on current density (Eq. (1)).
For a given cell configuration, with specified geometry and
material properties, as well as operating conditions (e.g.
pressure, temperature, and fuel and oxidizer composition),
each of the overpotentials must be evaluated as a function of
current density. Since the overpotential is usually an implicit
function of current density an iteration is needed. Most of the
iterations are accomplished using a NewtonRaphson
method. The solution procedure follows several steps as
follows:
Calculate the stoichiometric coefficients of the globalelectrochemical reaction by solving element balance
equations. This depends on the specific fuel and oxidizer
composition.
Based on the stoichiometric coefficients, evaluate thecharge transferred ne and the Nernst potential E.
Evaluate the species molar flux through the anode and thecathode (N
ga;k and N
gc;k). These are all evaluated explicitly
from the current density and the stoichiometry (Eqs. (23)
and (24)).
Calculate the species compositions at the electrodeelectrolyte interfaces ( Xk
s) from the DGM at given
species molar flux and species compositions in the chan-
nels. This is an iterative procedure.
Evaluate the concentration overpotentials for both elec-trodes (Zconc;a and Zconc;c) once the interface concentra-
68 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
9/14
tions are known (Eqs. (15) and (16)). This is an explicit
evaluation.
With a specified exchange current density i0 at the anodeand cathode, invert the ButlerVolmer equation to obtain
the activation overpotentials Zact;a and Zact;c. This is an
iterative procedure.
Calculate the ohmic overpotential Zohm in the electrolyte(Eq. (32)). This is an explicit procedure.
Determine the leakage overpotential Zleakage (Eq. (35)).Assuming that Z0leakage is specified, the value of imax must
be determined iteratively. The maximum current density
is achieved when the net voltage is zero. The value ofimaxdepends on the cell parameters as well as all the other
overpotentials. For a given current density ie, evaluate the
leakage overpotential.
With all the overpotentials in hand, evaluate the cellvoltage Ecell.
The procedure described above assumes that the cell
voltage is to be determined once a current density is
specified. To create a voltagecurrent plot, for example,
the procedure is repeated for a range of current densities.
When the MEA model is incorporated into a larger system-
level fuel-cell model, it is usually more natural to specify the
cell operating voltage and determine the local current den-
sity. In this case an outer iteration is implemented around the
whole procedure outlined above. Despite the nested itera-
tions, the convergence is very rapid.
4. Analysis of a methane-fueled SOFC
The objective of this section is to apply the model to a
particular anode-supported SOFC system operating with
methane as the fuel. Table 1 lists model parameters of the
unit cell. For an anode-supported SOFC system, both the
electrolyte and the cathode are much thinner than the anode.
Thus, it may be anticipated that the electrolyte, ohmic and
cathode-concentration overpotentials will be relatively
small. The transport parameters for calculating the ohmic
overpotentials in the electrolyte (Table 1) are based on the
oxygen-vacancy conductivity [28].
Assuming pure CH4 as a fuel and air as the oxidizer, Fig. 4
illustrates the predicted cell voltage and the power density as
a function of the current density. The peak power is seen
to be about 0:4 W/cm2 at about 0:4 V. Fig. 4 also showsthe contributions of all the overpotentials as shaded regions.
For this particular system, the cathode activation represents
the largest overpotential. It is followed by the anode-activa-
tion, ohmic, and anode-concentration overpotentials. The
cathode-concentration overpotential is so small that it does
not show on the plot. All the overpotentials increase with
increasing current density. The sum of the overpotentials
reaches the Nernst potential (in this case, about 1:15 V) atthe highest current density (ie % 2:25 A/cm
2) where the
power is reduced to zero. In other words, the cell ceases
to produce power when the overpotentials just equal the cell
potential.
4.0.1. Fuel utilization
Because fuel is consumed by electrochemical reaction as
it flows through the anode channel, the fuel stream is diluted
by product species (CO2 and H2O). Furthermore, as fuel is
Table 1
Parameters to represent a CH4 SOFC system
Parameter Value
Operating temperature (T, C) 750
Operating pressure (p, atm) 1
Anode
Thickness (La, mm) 1000
Porosity (fg) 0.35
Tortuosity (tg) 3.50
Average pore radius (rp, mm) 1
Average particle diameter (dp, mm) 10
Exchange current density (i0, A/cm2) 0.40
Charge-transfer coefficient (aa) 0.50
Number of electrons (nBVe ) 1
Cathode
Thickness (Lc, mm) 50
Porosity (fg) 0.35
Tortuosity (tg) 3.50
Average pore radius (rp, mm) 1
Average particle diameter (dp, mm) 10
Exchange current density (i0, A/cm2
) 0.13Charge-transfer coefficient (aa) 0.50
Number of electrons (nBVe ) 1
Electrolyte
Thickness (Lel, mm) 20
Activation energy of O2 (Eel, J/mol) 8.0E4Pre-factor of O2 (s0, S/cm) 3.6E5Leakage overpotential (Zmax, V) 0.0
Fig. 4. Cell voltage, overpotentials, and power density as a function of the
current density for CH4 as a fuel in an SOFC with parameters stated in
Table 1.
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 69
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
10/14
used, the Nernst potential decreases (Fig. 3). The various
overpotentials are also affected by the fuel depletion. Fig. 5
illustrates the cell performance for situations where the fuel
is 50, 75, and 85% used. In this example, the exchange
current densities i0 at the anode and cathode are set to be
constants (Table 1). Furthermore, the cathode channel is
presumed to be pure air (i.e. no oxygen depletion). At 75 and
85% fuel utilization there are seen to be critical current
densities at which the power abruptly terminates. In these
cases (approximately 1:57 A/cm2 for the 75% case andapproximately 0:88 A/cm2 for the 85% case), the anode-
concentration overpotential becomes so large that no fuelcan reach the anode three-phase boundary. As a result the
cell cannot operate.
The cell represented by Fig. 5 would operate well at
very high fuel utilization. For example, if the operating
voltage was set at 0:5 V the current density would varyrelatively little (approximately between 0:6 a n d 0:8 A/cm2) with corresponding variations in power density
(approximately between 0:42 and 0:32 W/cm2) as thecomposition in the fuel channel varies between pure
methane and a highly diluted stream where the fuel is
85% depleted.
Fig. 6 presents the result of another simulation where all
parameters except one are the same as for the simulation
shown in Fig. 5. The cell represented by Fig. 6 presumes that
the anode exchange current density is first order in the fuel
concentration. That is:
i0 i00XCH4 ; (36)
where XCH4 is the fuel mole fraction and i00 0:65 A/cm
2.
As the fuel is depleted, the exchange current density drops
proportionally. In this case, the results are strikingly differ-
ent. As the fuel is depleted the current density and the power
density are significantly diminished. The anode-activation
overpotential dominates the losses. It would likely be
impractical to operate such a cell at a fuel utilization greater
than about 50%.
5. Interpreting experimental observations
Gorte et al. [1,24] have measured SOFC performance for a
variety of fuels using a YSZ electrolyte and a CuceriaYSZ
cermet anode. Fig. 7 reproduces some of the reported
measurements together with results from our model.
Table 2 lists parameters used in our MEA model to fit
the experimental data for the case of H2 as the fuel. Some ofthe physical and operating parameters are reported [1,24],
but we have estimated others. The parameters at the cathode
are chosen such that the concentration overpotential is
very small. The transport parameters for calculating the
Fig. 5. Cell voltages and power densities as a function of the current
density for fuel CH4 at a different percentage of fuel utilization. The
exchange current density is set to be constant.
Fig. 6. Cell voltages and power densities as a function of the current
density for CH4 at a different percentage of fuel utilization. The exchange
current density is assumed to be the first order in the fuel concentration.
Fig. 7. The MEA model is used to represent a range of experimental data
of currentvoltage and power densities for an SOFC from Gorte et al.
[1,24] by varying a few parameters.
70 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
11/14
ohmic overpotentials in the electrolyte are based on the
oxygen-vacancy conductivity of YSZ [28]. To match the
experimental data for cases of CH4 and C4H10, the para-meters: porosity, exchange current density, reaction orders
and the leakage overpotential are adjusted as indicated in
Table 3.
As illustrated in Fig. 7, the experimental measurements
can be represented very well by our model through varying a
few parameters. We must hasten to caution, however, that we
claim no uniqueness in the set of parameters used. Moreover,
although the model can match the data, it is difficult to
support some of the best-fit parameters on physical
grounds. Nevertheless, we believe that one can gain valuable
insights through interpreting experimental observations in
the context of a quantitative, physically based model.
As shown in Fig. 7, the measured open-circuit cell
potential for the H2 case is very close to the theoretical
Nernst potential. However, for the CH4 and C4H10 cases the
measured open-circuit potentials are approximately 0.2 V
lower than the corresponding theoretical Nernst potentials.
Therefore, we have assigned a leakage overpotential to
capture the measured open-circuit potential (Table 3).
Close inspection reveals that the currentvoltage curves
for these three fuels have substantially different functional
forms. The varying shapes infer that the overpotential lossesfor these three fuels are dominated by different physico-
chemical processes. For the H2 fuel, the nearly straight-line
shape (for the entire range of current densities), indicates
that the major polarization may be due to the ohmic loss at
the electrolyte rather than the activation overpotential and
concentration overpotential at the anode. For CH4, the data
shows a substantial upward curvature at low current density
but a nearly linear function at high current density. In this
case, the major overpotential loss may be caused by the
relatively low anode activation. In the case of C4H10, the
currentvoltage curve is again functionally different from
the other fuels, particularly at the high current density. The
relatively lower power density indicates that the activation
overpotential is very important, and the downward curvature
at the high current density indicates a significant anode-
concentration overpotential.
Since the concentration overpotentials appear to be rela-
tively small for cases of H2 and CH4, we have used the
measured values for porosity and pore size. In these cases,
we have varied the exchange current densities and reaction
orders to fit these measurements. However, for the case of
C4H10 fuel, the anode-concentration overpotential appears
to dominate the potential losses, particularly at the high
current density. Consequently, it appears that the gas-trans-
port resistance through the anode must be increased. Themost direct way to increase transport resistance is to reduce
the anode porosity. A very low porosity off % 2% is neededto reproduce the observed strong curvature in the data. Since
the measured porosity is f % 50%, one must seriouslyquestion the physical validity of the porosity needed to fit
the data. It is more likely that there may be some important
physical process that is not represented in the model as it
currently stands. For example, is it possible that there is
significant catalytic reaction within the anode structure? If
so, perhaps it is not the parent butane fuel that reacts
electrochemically, but rather some reaction products. Cer-
tainly if one of the essential underpinnings of the model (e.g.
direct electrochemical oxidation of the fuel) is incorrect,
then inferences from the data-fitting procedure must be
questioned.
Table 2
Parameters for matching the H2 experimental data by Gorte et al. [1,24]
Parameter Value
Operating temperature (T, C) 700
Operating pressure (p, atm) 1
Anode (CuceriaYSZ)
Thickness (La, mm) 400
Porosity (fg) 0.52
Tortuosity (tg) 3.50
Average pore radius (rp, mm) 1.0
Average particle diameter (dp, mm) 3.0
Exchange current density (i00, A/cm2) 0.0055
Charge-transfer coefficient (aa) 0.50
Number of electrons (ne) 1
Cathode (LSMYSZ)
Thickness (Lc, mm) 50
Porosity (fg) 0.35
Tortuosity (tg) 3.50
Average pore radius (rp, mm) 1.0
Average particle diameter (dp, mm) 3.0
Exchange current density (i00, A/cm
2
) 0.75Reaction order of O2 0.5
Charge-transfer coefficient (aa) 0.5
Number of electrons (ne) 1
Electrolyte (YSZ)
Thickness (Lel, mm) 60
Activation energy of O2 (Eel, J/mol) 8.0E4Pre-factor of O2 (s0, S/cm) 3.6E5Leakage overpotential (Zmax, V) 0.07
Table 3
Parameters adjusted to match the experiment data by Gorte et al. [1,24]
Fuel type Reaction order (gk) f i00 (A/cm
2) Z0leakage (V)
H2 CH4 C4H10 H2O CO2
Hydrogen (H2) 1.0 0.0 0.0 1:0 0.0 0.520 0.0055 0.07Methane (CH4) 0.0 1.0 0.0 0:5 0.0 0.520 0.0045 0.20Butane (C4H10 0.0 0.0 1.0 1:0 0.0 0.021 0.0090 0.23
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 71
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
12/14
6. Summary and conclusions
We have presented a physically based mathematical model
that represents a unit cell of a fuel-cell membrane-electrode
assembly. It is written to accommodate general descriptions
of fuel and oxidizer mixtures. It considers a variety of anode
and cathode overpotential losses that include activation,concentration, ohmic, and leakage. The activation overpo-
tentials are described in terms of the ButlerVolmer equation.
Concentration overpotentials in the porous electrode struc-
tures are described in terms of a dusty-gas model. Ohmic
losses may stem from ion resistance in the electrolyte or
electrical resistance at material interfaces.
The model can be exercised to produce voltagecurrent
relationships for a certain cell configurations and specified
operating conditions. These simulations are valuable in
helping to understand the competing physical processes that
are responsible for controlling cell performance. Such
understanding can assist in cell design and optimization
as well as interpreting experimental observations.
The objective of this paper is primarily to describe the
MEA model and the underlying assumptions. The model can
be used as a stand-alone model. However, the software is
written to be incorporated as a submodel into larger system-
level models that also consider the fluid flow and thermal
transport for a full fuel-cell system.
The model presented here assumes direct electrochemical
oxidation of the fuel species. Thus, while there is multi-
component transport of fuel, oxidizer, and product species
within the electrode structures, there is no homogeneous or
catalytic reaction among these species. Although there is
reported experimental evidence of DECO [24] for hydro-carbon fuels in certain SOFC systems, there is also reason to
anticipate that catalytic reaction could occur within porous
cermet electrode structures. Thus, in future work we intend
to extend the models to accommodate this possibility.
Acknowledgements
This work has been supported by DAPRA through the
Palm Power program and by the DoD Multidisciplinary
University Research Initiative (MURI) program adminis-
tered by the Office of Naval Research under Grant N00014-
02-1-0665. We gratefully acknowledge frequent and sub-
stantial collaboration with Prof. Greg Jackson (University of
Maryland) and Prof. David Goodwin (Caltech). We also
acknowledge close and beneficial collaborations with grad-
uate student Kevin Walters and colleagues at ITN Energy
Systems, especially Neal Sullivan and Bill Barker.
Appendix A
This paper develops and discusses a general formalism
beginning with a global electrochemical reaction for direct
electrochemical oxidation. To maintain generality there is a
certain level of abstract nomenclature that must be used. The
objective of this appendix is to assist understanding and
interpreting the nomenclature by working through specific
examples.
A.1. C4H10air SOFC
The overall electrochemical reaction is stated as:
C4H10 13
2 0:210:21O2 0:79N2
! 4CO2 5H2O 13 0:79
2 0:21N2; (A.1)
which combines the electrochemical reaction in the anode:
C4H10 13O2 ! 4CO2 5H2O 26e
; (A.2)
and the electrochemical reaction in the cathode:
132 O2 26e ! 13O2: (A.3)
The fuel and oxidizer compositions are specified by setting
the non-zero nf;k and no;k as:
nf;C4H10 1; (A.4)
no;O2 0:21; no;N2 0:79: (A.5)
Balancing the reaction yields the non-zero stoichiometric
coefficients as:
nf0 1; (A.6)
no0
13
2 0:21
; (A.7)
n00f;CO2 4; n00f;H2O
5; (A.8)
n00o;N2 13 0:79
2 0:21: (A.9)
The non-zero charge transfer per mole of each reactant
species are:
zf;CH4 26; zo;O2 4: (A.10)
With the reaction balanced and the charge transfers
assigned, the number of electrons transferred by the global
reaction follows from Eq. (4) as:
ne XKk1
nf0nf;kzf;k 26: (A.11)
The concentration overpotentials are expressed as:
Zconc;a RT
26Fln
C4H10
C4H10s
4 ln
CO2
CO2s
5 lnH2O
H2Os
; (A.12)
Zconc;c RT
4Fln
O2
O2s
: (A.13)
72 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
13/14
For the butaneair case, the species molar fluxes in the anode
are:
NgC4H10
ie
26F; N
gH2O
5ie
4F; N
gCO2
4ie
8F;
(A.14)
and in the cathode are:
NgO2
ie
4F; N
gN2
0: (A.15)
A.2. Fuel-mixtureair SOFC
This example considers the anode channel to have a
mixture of fuel (CH4), products (H2O and CO2), and oxygen
(O2). At some particular point in the fuel-cell system the
mole ratios in the anode channel are presumed to be
CH4:H2O:CO2:O21.0:0.3:0.15:0.05.This oxygen is explicitly added to the anode streamit is
not coming through the MEA. The oxidizer is presumed tobe air. The overall electrochemical reaction can be repre-
sented as follows:
CH4 0:3H2O 0:15CO2 0:05O2
20:21 0:21O2 0:79N2
! 2:3H2O 1:15CO2 0:05O2 2
0:210:79N2:
(A.16)
The half-cell reactions combine the electrochemical reaction
in the anode:
CH4 4O2 ! CO2 2H2O 8e
; (A.17)
and the electrochemical reaction in the cathode:
2O2 8e ! 4O2: (A.18)
The fuel and oxidizer compositions are specified by setting
the non-zero nf;k and no;k as:
nf;CH4 1; nf;H2O 0:3; nf;CO2 0:15;
nf;O2 0:05; (A.19)
no;O2 0:21; no;N2 0:79: (A.20)
Balancing the reaction yields the non-zero stoichiometric
coefficients as:
nf0 1; no
0 20:21 ; (A.21)
n00f;CO2 1:15; nf;H2 O00 2:3; n00f;O2 0:05;
n00o;N2 2 0:79
0:21: (A.22)
The non-zero charge transfer per mole of each reactant
species are:
zf;CH4 8; zo;O2 4: (A.23)
With the reaction balanced and the charge transfers
assigned, the number of electrons transferred by the global
reaction follows from Eq. (4) as:
ne XKk1
nf0nf;kzf;k 8: (A.24)
The concentration polarizations at the anode and cathode are
expressed as:
Zconc;a RT
8Fln
CH4
CH4s
ln
CO2
CO2s
2 lnH2O
H2Os
; (A.25)
Zconc;c RT
4Fln
O2
O2s
: (A.26)
The species molar fluxes in the anode are:
NgCH4
ie
8F; N
gH2O
ie
4F;
NgCO2 ie8F;NgO2 0; (A.27)
and in the cathode are:
NgO2
ie
4F; N
gN2
0: (A.28)
References
[1] R.J. Gorte, S. Park, J.M. Vohs, C. Wan, Anodes for direct oxidation
of dry hydrocarbons in a solid-oxide fuel cell, Adv. Mater. 12 (2000)
14651469.
[2] E.P. Murray, T. Tsai, S.A. Barnett, A direct-methane fuel cell with aceria-based anode, Nature 400 (1999) 651659.
[3] S. Park, J.M. Vohs, R.J. Gorte, Direct oxidation of hydrocarbons in a
solid-oxide fuel cell, Nature 404 (2000) 265266.
[4] C. Wang, W.L. Worrell, S. Park, J.M. Vohs, R.J. Gorte, Fabrication
and performance of thin-film YSZ solid oxide fuel cells, J.
Electrochem. Soc. 148 (2001) 864868.
[5] A. Bjeberle, The electrochemistry of solid oxide fuel cell anodes:
experiments, modeling, and simulations, Ph.D. thesis, Swiss Federal
Institute of Technology, Zurich, 2000.
[6] A. Mitterdorfer, Identification of the oxygen reduction at cathodes of
solid oxide fuel cells, Ph.D. thesis, Swiss Federal Institute of
Technology, Zurich, 1997.
[7] A.S. Ioselevich, A.A. Kornyshev, Phenomenological theory of solid
oxide fuel cell anode, Fuel Cells 1 (2001) 40 65.
[8] S.H. Chan, K.A. Khor, Z.T. Xia, A complete polarization model of asolid oxide fuel cell and its sensitivity to the change of cell
component thickness, J. Power Sources 93 (2001) 130140.
[9] J.W. Kim, A.V. Virkar, K.Z. Fung, K. Mehta, S.C. Singhal,
Polarization effects in intermediate temperature, anode-supported
solid oxide fuel cells, J. Electrochem. Soc. 146 (1999) 6978.
[10] A.V. Virkar, J.Chen, C.W. Tanner, J.W. Kim, The role of electrode
microstructure on activation and concentration polarizations in solid
oxide fuel cells, Solid State Ionics 131 (2000) 189198.
[11] C.W. Tanner, K.Z. Fung, A.V. Virkar, The effect of porous composite
electrode structure on solid oxide fuel cell performance, I.
Theoretical analysis, J. Electrochem. Soc. 144 (1997) 2130.
[12] S.B. Adler, J.A. Lane, B.C.H. Steele, Electrode kinetics of porous
mixed-conducting oxygen electrodes, J. Electrochem. Soc. 143
(1996) 35543564.
H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174 73
-
8/8/2019 A General Mathematical Model for Analyzing the Performance
14/14
[13] J.R. Macdonald, Impedance Spectroscopy: Emphasizing Solid
Materials and Systems, Wiley, New York, 1987.
[14] R.J. Kee, F. Rupley, J.A. Miller, The CHEMKIN thermodynamic
database, Technical Report SAND87-8215, Sandia National Labora-
tories, 1987.
[15] E.A. Mason, A.P. Malinauskas, Gas Transport in Porous Media: The
Dusty-Gas Model, Elsevier, New York, 1983.
[16] T. Thampan, S. Malhotra, H. Tang, R. Datta, Modeling of conductive
transport in proton-exchange membranes for fuel cell, J. Electro-
chem. Soc. 147 (2000) 32423250.
[17] T. Thampan, S. Malhotra, J. Zhang, R. Datta, PEM fuel cell as a
membrane reactor, Catal. Today 67 (2001) 1532.
[18] R.J. Kee, G. Dixon-Lewis, J. Warnatz, M.E. Coltrin, J.A. Miller, A
Fortran computer code package for the evaluation of gas-phase
multicomponent transport properties, Technical Report SAND 86-
8246, Sandia National Laboratories, 1986.
[19] J. Bear, Dynamics of Fluids in Porous Media, Elsevier, New York,
1972.
[20] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals
and Applications, 2nd ed., Wiley, New York, 2000.
[21] H. Uchida, M. Yoshida, M. Watanabe, Effect of ionic conductivity of
zirconia electrolytes on the polarization behavior of various cathodes
in solid fuel cells, J. Electrochem. Soc. 146 (1999) 17.
[22] J.D. Kim, G.D. Kim, J.W. Moon, Y. Park, W.H. Lee, KI. Kobayashi,
M. Nagai, C.E. Kim, Characterization of LSMYSZ composite
electrode by ac impedance spectroscopy, Solid State Ionics 143
(2001) 379389.
[23] C.G. Vayenas, S.I. Bebelis, I.V. Yentekakis, S.N. Neophytides,
Electrocatalysis and electrochemical reactors, in: P.J. Gellings,
H.J.M. Bouwmeester (Eds.), The CRC Handbook: Solid State
Electrochemistry, CRC Press, Boca Raton, 1997, pp. 445480.
[24] R.J. Gorte, H. Kim, J.M. Vohs, Novel SOFC andoes for the direct
electrochemical oxidation of hydrocarbon, J. Power Sources 106
(2002) 1015.
[25] B. de Boer, SOFC anode: hydrogen oxidation at porous nickel and
nickel/yttria-stabilised zirconia cermet electrodes, Ph.D. thesis,
University of Twente, The Netherlands, 1998.
[26] M. Ihara, T. Kusano, C. Yokoyama, Competitive adsorption reaction
mechanism of Ni/yttria-stabilized zirconia cermet anodes in H2H2O
solid oxide fuel cells, J. Electrochem. Soc. 148 (2001) A209A219.
[27] X. Wang, N. Nakagawa, K. Kato, Anodic polarization related to the
ionic conductivity of zirconia at Nizirconia/zirconia electrodes, J.
Electrochem. Soc. 148 (2001) A565A569.
[28] K. Sasaki, J. Maier, Re-analysis of defect equilibria and transport
parameters in Y2O3-stabilized ZrO2 using EPR and optical relaxa-
tion, Solid State Ionics 134 (2000) 303321.
74 H. Zhu, R.J. Kee / Journal of Power Sources 117 (2003) 6174